A little less on your plate.

Folder of creatives → full paused campaign in 9 steps.

A launch that dies halfway picks up where it stopped — no duplicates.

3 sizes of one creative → one ad, each size in its own placement.

One ad, N languages, verified per locale.

Copy a campaign into another account — media re-uploaded, pixel and page swapped.

Country exclusions in 5 layers: global, funnel, language, fraud.

Pixel per funnel — exact match or loud error.

Names with machine-readable tags and a global number checked against Meta.

Funnel, language and producer — from file and folder names.

Which creative packs are ready to test — and in what order.

Creates N test campaigns a day, sized to free budget.

Images and videos of any size: chunks, readiness wait, reuse.

Give a tired ad set proven creatives of the same concept.

Creatives with great ROAS that run nowhere right now.

Alerts only on NEW disapproved ads, with the reason.

Creatives that spend and don't pay back — judged on closed cohorts.

Weekly: new winners, burned-out creatives, stale campaigns.

Top statics → why they worked → variations → QA → paused launch.

Kills losers, scales winners and rolls back bad raises on test ad sets — every night, with reasons.

A recommendation for every ad set, with a reason — applied by a button.

How much you can pay per trial, per funnel × language.

Every rule number in one place — bounds, audit log, contradiction check.

Every automated decision explained in plain words; doubtful ones flagged.

How often do your rules guess right: correct / wrong / unknown.

Replay rules on history before shipping a change.

Countries eating budget at poor ROAS — inside the campaign's language.

Keeps total spend under a ceiling — zones, hysteresis, weekday correction.

Account limit via +$500 / +$1k buttons — knows the dollars-vs-cents trap.

'Your credit line lasts N more days' — and an alert when it's cut.

One line blocks every new launch in a BM.

One door for alerts: what, why, what was done, what you need to do.

Pixel, CAPI and landing pages vs the same hour over 7 days.

No decisions until yesterday's data has landed.

Pixel purchases vs warehouse trials per funnel × language.

Did what you switched off actually stop spending?

Profitable campaigns someone switched off.

Spend, trials, CPT, ROAS by channel, funnel and language.

Owner's morning summary: totals, kill candidates, what changed.

Concept, format, author, age × real purchases.

Negative comments → creative risk score 1–100.

Competitors' hooks, visuals and funnels from Ad Library.

Blocks your agent from writing to Meta directly — only through recipes with guardrails.

Any creation not PAUSED is blocked.

Date, rules and freezes printed at the start of every session.

Before 'done': prove it, and name what wasn't checked.

Research → Summary → Implement paused → Verify. Verify can't be skipped.

Every mistake becomes one rule the agent reads next session.